brugada.net · The idea
Updated 16 July 2026
Not a cure. A hypothesis — and one with a specific, testable shape.
If the problem is a lost positive charge, the idea is almost insultingly simple: find something small and positive to stand in its place. A molecule that slips into the gap, grips the stranded negative charge, and holds the fold together long enough for the cell to finish building the protein and install it.
Think of a splint, not a repair. You’re not fixing the chain; you’re holding it steady while it does its own job. Drugs that work this way exist and have a name — pharmacological chaperones. They don’t block anything or switch anything on. They just help a protein fold.
There is precedent in this exact protein. A different Nav1.5 variant with the same kind of trafficking failure was rescued about 93-fold by an existing drug, mexiletine, in 2004. That’s the proof the category can work. Mexiletine itself was tested here and doesn’t bind this site — right idea, wrong pocket.
This is the part of the argument that is an idea. Everything on the previous page is measurement. This is a guess about what a molecule could do, and so far no molecule has done it anywhere except inside a computer.
A control experiment did something unexpected and useful. If you take the negative half of the clip and neutralise it too — so neither half is charged, and there’s no clip at all — the fold gets better, not worse.
That sounds backwards until you see what it means: the damage isn’t the missing bond. The damage is the abandoned charge. A buried electrical charge with no partner is expensive for a protein to carry around. Take the charge away and the problem goes with it.
Which reframes the target. A candidate molecule doesn’t need to recreate the original grip. It needs to neutralise a stranded charge — and it has to still be there hours later, not just at the moment of introduction.
There’s a related trap the project fell into and had to climb out of. The obvious way to score a candidate is “how often does it touch the target?” But the rescued control has the lowest touch-time of anything tested, and it works best. What matters isn’t contact — it’s whether the fold stops wobbling. Measuring the easy thing would have ranked everything exactly wrong.
Three 100 ns simulations were compared: wild-type, R104Q, and R104Q carrying the second-site D84N mutation that neutralises the stranded charge.
The D84N-rescued form recovers wild-type-like backbone flexibility (RMSF) — and it does so while having the lowest salt-bridge occupancy of all three. The bond forms less often; the fold is steadier anyway.
The distances go the “wrong” way too: R104–D84 measures 2.65 Å in wild-type, 3.21 Å in the mutant, and 3.76 Å in the rescued form. The rescue loosens the pair and still works, because the fix is charge elimination, not a tighter bond.
ESM-2 rates D84N mild/tolerated (LLR −0.58), which is why it works as a control rather than a second disease variant.
Why this matters for scoring: rank candidates by how often they touch D84 and you will rank them exactly wrong. The axis is rigidity, not contact.
08_cure_hypothesis · fig_rescue_mechanism · fig3_md_mechanism
The obvious alternative — use one of the many known drugs that act on this channel’s main pore — was tested and ruled out. The broken fold sits about 53 Å away from the pore, and the mechanical connection between them measures roughly 4,000× weaker than a genuinely connected pair of positions. You cannot reach this problem from over there. Push on the pore and nothing happens at the clip.
Decision DR-002. Signal transmission from the pore’s known drug site to R104 measures 3.9 × 10−5 — roughly 4,000× weaker than a genuinely coupled residue pair. The N-terminal domain sits ~53 Å from the pore.
A static pocket-druggability score near D84/R104 looked promising at 0.81 — and did not survive dynamics, dropping to 0.29–0.44. The site is real but shallow and soft, which is precisely why the strategy is a specific charge substitution rather than generic pocket-filling.
Mexiletine precedent: rescued the trafficking-deficient Nav1.5 mutant G1743R ~93-fold (Valdivia 2004, PMID 15023552). Tested here; does not bind this pocket; relegated to a reference control.
Osmolytes — the generic folding helpers (TMAO, betaine, taurine, proline, sarcosine) — did not repair this specific clip either.
Decision DR-002 · fig_pore_ntd_coupling · 05_docking_and_pockets
onwards.